Artificial intelligence used in mating planning by Pedigrees 360

Artificial intelligence (AI) and machine learning are at the heart of a new family tree analysis website launched by G1 Goldmine founder Leo Tsatsaronis. The Pedigrees 360 software is designed to help predict the likely success of matings by focusing on “line breeding clusters”.

“By using AI and machine learning, the program has demonstrated a very high level of accuracy in predicting which pedigree combinations will produce the champions and which will produce the eternal damsels,” said Tsatsaronis.

“Most existing mating tools focus on notches and crosses, which is a simplified representation of how a pedigree pattern affects a horse’s class. Clusters for deep line breeding are the basic pattern of the most successful breed-defining stallions of the last 100 years. Some of the greatest breeding heads in history, such as [Federico] Tesio, the Aga Khan, [Marcel] Boussac, designed their family trees to build multiple clusters and multiple layers of key ancestors going back as far as eight and nine generations.”

He continued, “Most of these great breeders had the ability to ‘see’ deep line breeding patterns and their non-computer vision was extraordinary.”

The software design for Pedigrees 360 includes nine-generation pedigrees for thousands of horses across a range of abilities, with the goal of finding the traits that separate them. Tsatsaronis noted that it is equally important to recognize which mating patterns are unsuccessful, especially when those patterns involve fashionable pedigrees. He also points to Pedigrees 360’s ability to identify potential quality horses from matings considered outdated by the commercial market and cites Knicks Go, North America’s Horse of the Year for 2021, as an example of what the software can find.

READ :  Envisioning the Battery Data Genome, a Central Data Hub for Battery Innovation

“Although the horse is the result of a mating with a low-wage sire and was sold at auction for a low price, Pedigrees 360 shows that the mating had more potential than conventional wisdom would suggest,” said Tsatsaronis.

“AI modeling analyzed 48 identifiable clusters within nine generations, where clusters were items such as number of sire duplications, number of sex-balanced mare duplications in seven, eight and nine generations, and inbreeding positioning between generations.

“Of course there will always be horses that belie their ancestry. Some that the program says will be outstanding are actually slow and vice versa, but so far this has proven to be a small percentage and perhaps the result of a biomechanical factor rather than pedigree.”

He added: “When most breeders and buyers are evaluating a potential pairing or yearling, they consider a variety of factors including pedigree, conformation, x-rays and exercise. Our program is just a tool to help with one of those factors, but it is a very powerful tool.”

Hutton Goodman from Mt. Brilliant Farm is among the North American breeders who have signed up to Pedigrees 360. He will continue to sharpen and update it to make it more useful as he did with G1 Goldmine. We’ve already started to see some of the great features he’s developing to hopefully add and it’s exciting things.”

For more information on the launch of the new website, visit